THE APPEARANCE OF SPICULES IN HIGH RESOLUTION OBSERVATIONS OF Ca II H AND Hα

Tiago M. D. Pereira, Luc Rouppe van der Voort, and Mats Carlsson

Institute of Theoretical Astrophysics, University of Oslo, P.O. Box 1029 Blindern, NO-0315 Oslo, Norway; tiago.pereira@astro.uio.noReceived 2016 March 8; accepted 2016 April 8; published 2016 June 14

ABSTRACT

Solar spicules are chromospheric fibrils that appear everywhere on the Sun, yet their origin is not understood.Using high resolution observations of spicules obtained with the Swedish 1 m Solar Telescope, we aim tounderstand how spicules appear in filtergrams and Dopplergrams, how they compare in Ca II H and Hα filtergrams,and what can make them appear and disappear. We find that spicules display a rich and detailed spatial structure,and show a distribution of transverse velocities that, when aligned with the line of sight, can make them appear atdifferent Hα wing positions. They become more abundant at positions closer to the line core, reflecting adistribution of Doppler shifts and widths. In Hα width maps they stand out as bright features both on disk and offlimb, reflecting their large Doppler motions and possibly higher temperatures than in the typical Hα formationregion. Spicule lifetimes measured from narrowband images at only a few positions will be an underestimatebecause Doppler shifts can make them disappear prematurely from such images; for such cases, width maps are amore robust tool. In Hα and Ca II H filtergrams, off-limb spicules essentially have the same properties, appearance,and evolution. We find that the sudden appearance of spicules can be explained by Doppler shifts from theirtransverse motions, and does not require other convoluted explanations.

Key words: Sun: atmosphere – Sun: chromosphere

Supporting material: animation

1. INTRODUCTION

Spicules and fibrils are observed all over the Sun inchromospheric lines. Their very existence and transient, fastmotions have been a challenge to explain. Much work has beencarried out on this subject, with early reviews by Beckers(1968, 1972) and later reviews by Sterling (2000), Rutten(2012), and Tsiropoula et al. (2012). Some of the most pressingquestions about spicules are (1) what drives them and (2) whatis their contribution to the transfer of energy and mass from thephotosphere to the corona.

When coined by Roberts (1945), the term “spicules” appliedstrictly to objects outside the solar limb. Since then, manyobjects that are believed to be their disk counterparts have beenobserved on the solar disk, resulting in a profusion of differentterms used to refer to (mostly) the same objects (Beckers 1968;Grossmann-Doerth & Schmidt 1992; Tsiropoula et al. 1994;Rutten 2006; Langangen et al. 2008; Judge et al. 2012; Sekseet al. 2013b). Here we adopt the term “spicules” for both limband disk objects, clarifying whether they are “off limb” or “ondisk” when necessary.

What is the allure of spicules? What makes them aworthwhile research topic? For one, they appear nearlyeverywhere on the Sun and are dominant in some chromo-spheric filtergrams such as Ca II H. Physically resembling jets,rooted in the photosphere, and reaching coronal heights, fromearly on it was natural to assume that they may contributetoward heating the corona and supplying it with mass. Indeedearly estimates indicated that the upward mass flux of spiculescan be 100 times larger than the solar wind (Pneuman &Kopp 1977, 1978); even if most of that flux comes back down,if only a few a few percent continue upward, it is enough todrive the solar wind.

Much of the research on spicules has followed fromadvances in observations. Early hopes for their importancein coronal heating were dashed when no spicular emission

was observed in coronal lines (Withbroe 1983). Interest wasrekindled when Hinode observations revealed thatsome spicules (so-called type II) were more violent thanpreviously thought, fading from the Ca II H passband and notseen to fall back down (De Pontieu et al. 2007). While Zhanget al. (2012) questioned the existence of two types ofspicules, Pereira et al. (2012) analyzed some of the samedata sets and concluded otherwise, finding that type II arethe dominant type. The fading of Ca II H spicules has beenlinked to higher energy emission from the transition regionand corona (De Pontieu et al. 2009), and the advent of theInterface Region Imaging Spectrograph (IRIS, De Pontieuet al. 2014) has shown that spicules have a clear signal intransition region filters and they continue to evolve inhigher temperatures after fading from Ca II H (Pereira et al.2014). Judge & Carlsson (2010) suggest that despitetheir dominance in chromospheric images, spicules makeup less than 1% of the mass of the whole chromosphere.Judge et al. (2011) speculate that spicules may not be realmass motions, but optical illusions arising from warpedmagnetic sheets, a view that Judge et al. (2012) claim tofind observational evidence for. All of these findingshighlight the fact that the understanding and modeling ofspicules relies on their observed properties, some of whichare still contested.This work aims to contribute to the discussion of the

observed properties of spicules, in particular how differentobservations relate to one another and how to properly tracethe histories of spicules. To achieve those aims, we make useof a unique set of observations, which is introduced inSection 2. In Section 3 we compare spicules in Ca II H andHα, and in Section 4 we discuss their appearance infiltergrams and Dopplergrams. We discuss our findings inSection 5 and finish with a summary of our results inSection 6.

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 doi:10.3847/0004-637X/824/2/65© 2016. The American Astronomical Society. All rights reserved.

1

2. OBSERVATIONS

We obtained a series of observations at the Swedish 1 m SolarTelescope (SST; Scharmer et al. 2003) on La Palma using theCrisp Imaging SpectroPolarimeter (CRISP; Scharmeret al. 2008) instrument. CRISP is a dual Fabry–Pérotinterferometer (FPI) that contains three high-speed CCD cameras(35 frames s−1 with an exposure time of 17ms per frame): twocameras behind the FPI and a polarizing beam splitter, and athird “wide band” camera located before the FPI, which is usedas an anchor channel for image processing. The CRISP field ofview is approximately 61″ × 61″, with a plate scale of0 058 pixel−1. For this work we used a CRISP configurationscanning only the Hα line: using a pre-filter with a full width athalf maximum (FWHM) of 0.49 nm we scanned the line along25 positions, from −0.12 to 0.12 nm around the line core in0.01 nm steps. The FPI allows for very fast wavelength tuning(<50ms) within a spectral line; the cadence of our setup was5.5 s. In addition to CRISP, we used the so-called “blue tower”at the SST (see, e.g., Henriques 2012). Using a dichroic beamsplitter, the blue part of the spectrum (λ < 500 nm) waschanneled to two additional imaging cameras: one was placedbehind a Ca II H interference filter (FWHM of 0.11 nm, centeredat 396.88 nm) and the other behind a wide band filter (FWHM of1 nm) centered at 395.37 nm (the pseudo-continuum between theCa II H and Ca II K lines).

The observations took place on 2014 June 17; the timesequence analyzed here was obtained between 10:20 UT and11:15 UT. For this period the atmospheric conditions on LaPalma resulted in excellent, stable seeing. The target was quietSun at the solar north pole, centered at solar (x, y) coordinatesof 24″, 939″. In Figure 1 we show the target in context withimages from the Atmospheric Imaging Assembly (AIA; Lemenet al. 2012) in the 30.4 and 19.3 nm channels. As can be seen inthe 19.3 nm images, there was no polar coronal hole during thisperiod.

The CRISP data were reduced using the CRISPRED pipeline(de la Cruz Rodríguez et al. 2015). We made use of the Multi-Object, Multi-Frame Blind Deconvolution (MOMFBD) imagerestoration technique of van Noort et al. (2005), and employedthe cross-correlation method of Henriques (2012) to minimizethe seeing deformations introduced by the non-simultaneity ofthe narrowband CRISP images. The blue images were flatfielded and dark subtracted, and were also restored usingMOMFBD. The wide band cameras of CRISP and the bluebeam were used to co-align both series.

To enhance the visibity of spicules, we employed radialdensity filters (see discussion in Skogsrud et al. 2015, andreferences therein). These filtered images are built by dividingthe images by a mean intensity profile as a function distance tothe limb. For the CRISP images shown, we applied radial filtersfor each wavelength independently. Unless otherwise noted, allimages shown here have been radially filtered. Following onthe discussion of Skogsrud et al. (2015) we made sure that theradial filters left no spicular signal under the noise.

To visualize, connect, and interpret the data we madeextensive use of CRISPEX (Vissers & Rouppe van derVoort 2012).

3. COMPARING Hα WITH CA II H SPICULES

While spicules have been observed in many chromosphericlines, most studies in the early literature (pre-Hinode) made use

of Hα observations (see Beckers 1968, 1972). The advent ofHinode, with SOTʼs Broadband Filter Imager (BFI) high-quality Ca II H filtergrams led to renewed interest in spiculesand the discovery of a widespread new dynamic behavior (e.g.,De Pontieu et al. 2007; Pereira et al. 2012). When confrontingthe results from Hinodeʼs Ca II H spicules with earlier literature,one can ask why the more dynamic behavior of type II spiculeswas not observed in earlier Hα observations. Pereira et al.(2013) showed that by just degrading the Hinode data to similarconditions of earlier observations, one can derive the propertiesof classical spicules. The authors also compared SOT BFI Ca II

H filtergrams with NFI Hα wing filtergrams and concluded thatthe spicules were very similar in both lines. Nevertheless, slightdifferences in the spicules were caused by comparing acomposite Hα ± 80 pm wing filtergram with a broader filterin Ca II H. Such differences could, in an extreme scenario, stillcause doubt on whether the evolution of spicules would be thesame in both spectral lines.Here we compare Ca II H filtergrams with synthetic Hα

filtergrams computed from the CRISP spectra. In Figure 2 weshow a comparison between the Ca II H and Hα filtergrams fordifferent regions and times. The Hα filtergrams were calculatedby convolving the CRISP spectral images with a Gaussiantransmission function with a FWHM of 0.1 nm (the choice ofFWHM is discussed in the next paragraph). The Ca II H imagesshown suffer from fringing, which is an observational artifactthat was not possible to remove in post-processing because ofits time dependency. These fringes show up clearly as regularcircular or linear patterns, are enhanced by the radial filtering,and are noticeably different from the underlying spicules.A 0.1 nm Hα filtergram still shows a moderate amount of

opaque material just above the limb, with individual spiculesdifficult to discern. On the Ca II H filtergrams, the individualspicules can be traced all the way to the limb. The reason forthis apparent discrepancy is that for the same distance near theline core, Hα images still have a considerable amount ofchromospheric contribution, while in Ca II H the wings areprogressively formed much closer to the photosphere. Using awider filter for the Hα filtergrams, one can obtain nearlyidentical results. Our Hα wavelength window is only 0.24 nm,and therefore for widths larger than 0.1 nm our syntheticfiltergrams cannot sample the far wings of Hα. Nevertheless,we estimate that with a FWHM of 0.3 nm the Hα filtergramswill look closest to the 0.11 nm Ca II H filtergrams.Figure 2 shows a remarkable similarity between the Hα and

Ca II H spicules. Using arrows as a visual aid, we note severaldetails of fine structure and spicule length. Aside from smalldifferences in intensity and noise levels (and the fuzzier bottomhalf due to the Hα filter), we find that the spicule shapes,extent, and lifetimes are essentially identical between Hα andCa II H. In this quiet Sun region most of the spicules are of typeII, fading at around their maximum extent. The same isobserved in Hα. Long sequence movies show the sameevolution in both filters.

4. SPICULES AND FIBRILS IN FILTERGRAMS ANDDOPPLERGRAMS

4.1. Off limb and On disk

Spicules have been traditionally observed in Hα and Ca II Hfiltergrams (as shown above, the differences in spiculeproperties between both lines are negligible). Both broad and

2

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

narrow filtergrams centered in the cores of chromospheric linesshow little evidence of spicules on the solar disk—hence theywere classified earlier as limb objects. With broad filters thespicule signal is very low compared with the signal ofphotospheric light present in the line wings (Carlssonet al. 2007; Beck et al. 2013). With narrow filters individualspicules are usually indistinguishable from the crowded canopyof chromospheric fibrils. But spicules have clear spectralsignatures in the form of Doppler-shifted, wider line profileswith increased absorption in the red or blue wings. Suchsignatures have been observed for spicules on disk (e.g.,Langangen et al. 2008; Rouppe van der Voort et al. 2009;Sekse et al. 2013b; Yurchyshyn et al. 2013; Kuridzeet al. 2015) and off limb (Pereira et al. 2014).

In Figure 3 we show the same field of view in different Hαimages. In the top panel we show narrowband CRISP images atwavelengths from −50 to 50 km s−1 from the line center. At±50 km s−1 one can see the limb spicules very clearly, while

only a faint hint of the disk spicules is visible. This is likelybecause fainter events on disk will be obscured by thephotospheric light background, and a radial filter was appliedon the limb spicules. As one looks closer to the line core, thespicules become more abundant and their bundling as “bushes”rooted in the network becomes more obvious.In the bottom panel of Figure 3 we show different composite

quantities: five Dopplergrams, a synthetic filtergram, and a mapof normalized Hα widths. The Dopplergrams were built bytaking the difference of two CRISP narrowband images atsymmetric positions from the line center. Such differenceimages highlight regions with strong Doppler shifts. On disk,with Hα in absorption, negative velocities (blueshifts) appearblack, while positive velocities (redshifts) appear white. Offlimb, when Hα is in emission, the reverse is true: blueshiftsappear white, while redshifts appear black. When a featureappears in a Dopplergram at a particular velocity, it does notmean that it has a flow at that velocity—instead it means that its

Figure 1. Our observations in context. Top panels show images from AIA in the 30.4 and 19.3 nm filters. The CRISP field of view is represented by the cyan square.In the bottom panels we show an image from CRISP in the blue wing of Hα (−32 km s−1 from the line core) and an image from the Ca II H filter. Both Hα and Ca II Hhave been radially filtered to enhance the visibility of spicules (see the text). The observing time (in UT) of each image is printed in the top right corners.

3

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

line profile is Doppler shifted enough or wide enough so thatsome intensity decrease/increase takes place at that wavelength(see discussion in Lipartito et al. 2014). Spicules stand out insuch Dopplergrams, which eliminate most of the photosphericbackground and show in great spatial detail their predominantline of sight motions. The abundances of spicules at differentDopplergrams reflect their velocity distribution (of the threekinds of motions: upflow, transverse, and torsional; see Sekseet al. 2013b). In the far wing images one sees a lot less events,presumably the extreme tail of the motion distribution.However, as one moves farther from the limb, the Hα profilealso gets naturally narrower. Therefore, the same velocity shiftthat would cause a disk spicule to be visible in the far wingDopplergrams will not necessarily make it visible in the sameDopplergram above the limb. Hence we often see longer limbspicules at wavelengths closer to the line core (both in imagesand Dopplergrams). To compensate for this and increase thevisibility of spicules in Dopplergrams, we created a “compositeDopplergram” by combining two Dopplergrams made ofradially filtered images: one at ±50 km s−1 (±0.11 nm) andthe other at ±5 km s−1 (±0.01 nm), also shown in Figure 3.Both Dopplergrams were multiplied by a step function with asmooth profile: the lower part of the image is the ±50 km s−1

Dopplergram and the upper part is the ±5 km s−1 Doppler-gram; the dashed line is the middle point where they areblended. Thus the combined Dopplergrams are a “cleaner” wayto visualize spicules from the limb to their very top, mostlyavoiding the chromospheric “haze” that is prominent whenlooking just above the limb at wavelengths close to theline core.

From the synthetic filtergram one can see the advantage ofbroadband images over narrowband images: they capturesignals from spicules on both red and blue wings. Despitehaving a relatively narrow (0.1 nm) FWHM, the syntheticfiltergram still shows disk spicules as very faint and difficult to

identify, and the considerable opacity of chromospheric fibrilsalso makes it difficult to trace the lower part of limb spicules.Broader filters make the identification of limb spicules easier,but make limb spicules even fainter (e.g., Hinode SOT BFIʼsCa II H filtergrams).One key diagnostic in Figure 3 is the normalized Hα width.

For each pixel, the width W was calculated as the FWHM ofthe line profile. On the disk the line is in absorption, so this iswell defined. For locations more than ≈3Mm above the limb,the line is in pure emission, so the FWHM is also well defined.However, for regions in the first ≈3Mm above the limb itbecomes problematic to define a FWHM because the lineprofiles are transitioning from absorption to emission and willhave a variety of shapes in between. In such cases the emissionbegins as peaks on the far blue and red wings, which becomebigger away from the limb, going to an absorption peak with acentral reversal peak and finally pure emission. MeasuringW asFWHM in those intermediate profiles is often meaningless(e.g., when there is emission in the wings but an absorptioncore still deeper than the far wings, half of the maximum is anarbitrary location). Our approach to measuring W in alllocations was to start on the outside edges of the wings andmove inward toward the line core, finding the wavelengthdifference between the first positions where the intensity equalshalf of the maximum. This minimizes only some of the issuesin the intermediate limb region. Because our CRISP coverageonly goes ±0.12 nm around the line core, the very far wings ofHα are not covered and the maximum intensity of the spectralies in these far wings and not the continuum. To compensatefor all these limitations, we normalize the W by mW ( ) (i.e., thewidth of the mean spectrum at the same distance from thelimb), which was calculated by averaging the complete timeseries along radial curves parallel to the limb. Therefore thequantity plotted in Figure 3 can be seen as a “widthenhancement.” It is also visible that in the critical region up

Figure 2. Comparison of filtergrams in Hα (top row) and Ca II H (bottom row). The images are from 2014 June 17, the UT time is shown on the top right corners. Tohighlight common features and tops of spicules, red arrows were placed in the same positions in the top and bottom rows. Each red arrow has a length of 1 45 (1.05Mm). The size of each image is 24″ × 17″ (17.4 × 12.3 Mm2). The Hα filtergram was built by convolving the CRISP spectrograms with a Gaussian transmissionfunction, using a FWHM of 0.1 nm. The Ca II H filtergram has a FWHM of 0.11 nm.

4

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

to 1.7 Mm above the limb there are many cases where thewidth maps are very irregular—such regions should be ignored.Just above the spicules the width maps in Figure 3 appearwhite, meaning large width. This effect comes from the noise—the real signal being so weak that our approach mistakes noisefor the signal and finds large widths. We chose not to maskthese areas because these white areas above the spicules are anindicator of no spicule signal. The black areas just underneath,despite showing very low widths, are an indicator that there isstill an emission profile and a little spicule signal is left (e.g.,the large spicule at (x, y) ≈ (2, 18)Mm appears prominently innear core images and Dopplergrams, but the emission lineprofile is rather narrow, so it appears mostly as dark in thewidth map).

By taking the normalized width maps, we put the limbspicules on a common (width enhancement) scale as the diskspicules. They both appear, very noticeably, as widthenhancements. We find numerous spicules in bushes rootedin the network, and they stand out even more from thebackground than in Dopplergrams. As in broadband filter-grams, width maps allow one to see both redshifted and

blueshifted spicules, but with the huge advantage of beingsensitive to both spicules on the disk and above the limb(notwithstanding the issues at the limb noted above). Crucially,width maps are much more reliable at tracing the full lifetimeand length of spicules, which can prematurely disappear fromDopplergrams and filtergrams.

4.2. Appearance and Disappearance

The idea of spicules as jets of chromospheric material is asold as the term “spicules” itself (Roberts 1945). From earlyobservations (e.g., Roberts 1945; Lippincott 1957) the early lifeof spicules was observed as an apparent upward motion. Thisview has so far stood the test of time: most studies agree thatspicules show an upward motion in the early phase of theirlives (e.g., Beckers 1968, 1972; Nishikawa 1988; Suematsuet al. 1995; De Pontieu et al. 2004; Tsiropoula et al. 2012).What happens in the later stages was unclear early on(Beckers 1968), and from Hinode Ca II H filtergrams it appearsthat spicules can be divided into two distinct groups (DePontieu et al. 2007): type I spicules show a rise and fall, whiletype II spicules seem more violent and simply fade from Ca II H

Figure 3. Hα observations and properties of spicules and fibrils observed on 2014 June 17, 11:43 UT. Top panels: Hα intensities at different positions in the lineprofile, from −50 km s−1 to 50 km s−1 (labels shown at the top), all radially filtered to enhance spicule visibility. Bottom panels: from left to right, with labels shownat the top: Dopplergrams at different distances from ±50 to ±5 km s−1, filtergram with 0.1 nm Gaussian FWHM, Hα width divided by the mean width at each μvalue, and combined Dopplergram (see the text for details). Intensities in all panels were individually scaled for improved visibility and contrast.

(An animation of this figure is available.)

5

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

filtergrams with no downward motion observed (type IIspicules dominate in quiet Sun and coronal holes; see Pereiraet al. 2012). Analyses using IRIS data (Pereira et al. 2014;Skogsrud et al. 2015) show that after fading from Ca II Hfiltergrams, spicules continue to evolve and show a downwardphase in filtergrams sampling higher temperatures, suggestingrapid heating.

Despite mounting evidence pointing toward spicules asejections of material, some observational properties remaindifficult to reconcile with this view. Perhaps most important,the complete time evolution is observed for only a smallnumber of them. In movies many of the spicules simply appeararound their maximum length, with no clear observation oftheir rise or when they started. The swaying transverse motionof spicules in their dense bushes is a likely cause. The largespatial superposition of spicules, particularly at the limb, makesit challenging to track every individual spicule (e.g., see“Selection Effects and Errors” in Pereira et al. 2012). Spiculesalso recur frequently from the same footpoint region (Beck-ers 1968; Suematsu et al. 1995; De Pontieu et al. 2011; Pereiraet al. 2012; Sekse et al. 2013b; Yurchyshyn et al. 2013),meaning that when a particular spicule disappears it can bequickly replaced by another of similar intensity in the sameplace. Judge et al. (2012; hereafter JCR12) study disk spiculesat a wavelength of Hα + 0.11 nm, and find spicules that appearand disappear suddenly over several Mm. The authors claimthis is evidence that spicules are not jets but instead an opticalsuperposition caused by sheets of chromospheric material, aview earlier postulated by Judge et al. (2011). Our analysis is atodds with this interpretation.

We analyze a few cases of the so-called “suddenly appearingspicules” found by JCR12. In our data set those events seemrather rare: visual inspection finds only a handful of clear casesin the whole time series, and certainly a lot less than the 1/3–1/8 estimate of Lipartito et al. (2014, hereafter LJKG14).Here we note that the observations used by LJKG14and JCR12 were from an active region while we observe thequiet Sun. In Figure 4 we illustrate the time evolution of one ofthese events. In the top row we show CRISP images at+50 km s−1 or +0.11 nm away from the Hα core (red wing),which is the same wavelength used by JCR12. In the secondrow we show the Hα + 41 km s−1 (+0.1 nm) intensity and thenormalized width mW W ( ). On the bottom panel we showspectrograms along the spiculeʼs axis (shown as a dashed linein the upper three rows). The spectrograms were divided by themean spectrum as a function of viewing angle, which as notedby LJKG14 makes it easier for spicules to stand out: theyappear as dark bands the line wings. The normalized width(scale on top, from 0.9 to 1.25) along the spicule axis isoverplotted on the same panel. At t ≈ 22.1 s a spicule appearsseemingly along its whole length at +50 km s−1, and is visibleuntil at least t ≈ 60.8 s. At +41 km s−1 the spicule appearssooner, at t ≈ 11.1 s, and is visible for at least a frame longerthan at +50 km s−1. Finally, in the width images it is clear thata structure is already present before it is visible in any of the redwing images, and already at the start of the sequence shown(unfortunately, for previous times it is not easy to follow thisspiculeʼs evolution in the width maps). On the spectrogramsone can see how the width and Doppler shift of the featureevolve. In the first frame one can see a slightly increased width,

m »W W 1.05( ) in the first Mm of the structure. Thenormalized line profile shows a darkening in both wings. As

time evolves there is a clear increase in width and redshift.When the spicule appears at +50 km s−1, W/W(μ) ≈ 1.15, anda redshift is evident by the red wing being very dark and theblue wing being lighter and even appearing as white (meaningits intensity is above the mean spectrum intensity). At the peakof strongest wing absorption, the spicule has a widthenhancement of about 1.25, which later lowers as it evolves.In the last frames the redshift along the spicule is still visible, ahint that there could be different mechanisms for the widthenhancement and transverse motions. The example shown inFigure 4 is not atypical; all the suddenly appearing spicules weanalyzed were visible in width maps before appearing at Hα +50 km s−1 (but again, we stress that in our data set these eventswere rare).Figure 5 shows a different example of suddenly appearing

spicules, now at the limb. We compare Hα wing intensities at±23 km s−1 (±0.05 nm) from the line center with a combinedDopplergram (see above) and a synthetic filtergram with aFWHM of 0.1 nm. For compactness and the sake of clarity, thefirst frame already shows an intermediate stage in the life of thespicule, and only every second observed frame is shown. In theblue wing images at −23 km s−1 we see a large spicule risingabout ≈8Mm above the limb (and above the fibril canopy at≈5Mm). The spicule has multiple strands or threads. In thefirst frame its top half is noticeably blueshifted. From thecombined Dopplergram, especially at later stages, one can findwhat looks like the bottom part of the spicule (or at least astructure that is aligned with what one would expect to bebottom part of the spicule); the bottom part is also multi-stranded and shows an increasing redshift: negligible in the firstframe, then slowly increasing until t ≈ 55.3 s. Seen in the bluewing, the top of the spicule starts getting fainter until it quicklydisappears at t ≈ 44.3 s. On the other hand, the same structuregets progressively brighter in the red wing images, because it ishardly visible in the first frame and reaching maximumbrightness at around t ≈ 60 s. After t ≈ 66.4 s the spiculestarts getting fainter in the red wing and appears quickly in theblue wing images. This appearance and disappearance from thewing images is a consequence of line of sight velocity shiftsfrom the swaying motion of the spicule. This is made clearer inthe Dopplergrams, where the top of the spicule quickly goesfrom white (blueshifted) to black (redshifted) and again towhite. Different strands of the spicule also appear to be movingwith different velocities. Sensitive to both wings of the line, thefiltergram intensities show a rather consistent spicule intensitywith no sudden appearances or disappearances.The example in Figure 5 is a rare occurrence. For such a

scenario to be clearly observed, the spicule has to be taller thanmost and with a transverse motion closely aligned toward theobserver (for shorter spicules there is too much superposition).Nevertheless, this example highlights the limitations ofnarrowband images for studying the lives of spicules. By onlylooking at a particular wavelength one can miss part of theevolution, assume that the spicule appears or disappearssuddenly, and confuse different stages of the same event asdistinct, recurring events.

5. DISCUSSION

We showed how spicules appear in different positions in Hα,in filtergrams and Dopplergrams. Comparing limb spicules inCa II H and Hα filtergrams we find that there is very littledifference between the two filters. The same fine scale and

6

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

features are visible, and spicules evolve in the same way. Thisbeing a quiet Sun region, the overwhelming majority ofspicules are type II spicules, and they fade from bothfiltergrams at the same time. This is important for reconcilingthe results from Hinode with earlier work on spicules, whichwas mostly done using Hα observations. In the post-Hinodeliterature there are very few studies of type II spicules observedwith Hα. Pasachoff et al. (2009) most likely observed type IIspicules in Hα, but arrived at the puzzling conclusion that themajority of their spicules were not type I because more than70% faded and did not descend; but they were also not type IIspicules, because their upward apparent velocities were too low(close to the canonical value of 25 km s−1, e.g., Beckers 1968).However, with a cadence of ≈50 s, their study misses thedetailed dynamics and is most likely affected by spiculeconfusion due to superposition, which Pereira et al. (2013)found to underestimate the apparent velocities and overestimatethe lifetimes. In addition, Pasachoff et al. (2009) measure theapparent velocity as the mean velocity of the spicule in anumber of frames, whereas De Pontieu et al. (2007) and Pereira

et al. (2012) measure the upward velocity as the maximum (orstarting) velocity. Combining all these factors, one suspectsthat the velocities of Pasachoff et al. (2009) can probably bereconciled with typical velocities of type II spicules(30–110 km s−1), and that the authors did observe type IIspicules in Hα. This would confirm our findings that in Hα,type II spicules fade at the end of their lives, just like whenobserved in Ca II H filters.Spicules become more abundant and densely packed when

observed at wavelengths closer to the line core of Hα. Theappearance of absorption (or emission off the limb) in thewings of Hα is caused by both Doppler shifts and enhancedline widths (LJKG14 estimate that similar magnitude changesin either will change the wing intensities to a comparabledegree). Therefore, the decrease in spicule numbers whenobserving away from the line core to the wings reflects thedistributions of line shifts and widths; only the most extremeevents are seen in the outer wings. In narrowband filter imagesaway from the core (Dv∣ ∣ 30 km s−1) spicules stand out veryclearly against a mostly photospheric background, so such

Figure 4. Hα observations of a spicule/fibril on disk. The top three rows show, from top to bottom, the time evolution in the Hα red wing intensities at 50 km s−1

(0.11 nm) from line center, 41 km s−1 (0.09 nm) from line center, and W, the FWHM of Hα (see text). The spatial size of this small region is 1.5 × 3.5 Mm2. Thedotted red lines are cuts along the spicule axis, along which spectrograms are shown in the bottom row. The spectrograms have been divided by the mean Hα lineprofile, therefore black and white represent regions with lower and higher intensity than the mean. Superimposed over the spectrograms are plots of the line widthdivided by the mean width, with the scale on top (major tickmarks at 1 and 1.2). Time increases from left to right, and the images are shown at the observationalcadence of 5.5 s, as noted at the top.

7

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

images have been extensively used to measure the properties ofspicules on disk (e.g., Langangen et al. 2008; Rouppe van derVoort et al. 2009; Sekse et al. 2012; JCR12; Sekseet al. 2013a; LJKG14). However, as shown in Figure 4, thetransverse swaying motions can make a spicule appear shorter-lived when observed further from the line core. In the exampleof Figure 4, at 50 km s−1 the spicule lives for about 40 s, whilewhen taking the full spectral information into account, theeventʼs lifetime is almost twice as long. The full lifetime is notavailable from a single narrowband image because of changingDoppler shifts. This provides a natural explanation (alreadyspeculated by Sekse et al. 2013a) to the fact that lifetimes fordisk spicules measured from narrowband images (e.g., Sekseet al. 2013a) are about half the lifetimes for limb spicules(measured from broadband filtergrams, e.g., Pereiraet al. 2012).

We have shown that Hα width maps are much more reliablefor studying the dynamics of spicules than narrowband images.In such maps spicules stand above the background as brightfeatures and the full history is seen (i.e., not sensitive toDoppler shift changes). Cauzzi et al. (2009) computed similarwidth maps (calling them a “core width” because, as inour observations, the full spectral profile was not available)in the network, where some spicules are also seen standingout from the dark background. Increased Hα widths areassociated with macroscopic motions, but also to temperatureincreases (Leenaarts et al. 2012). Applying the relations

found by Leenaarts et al. (2012) to estimate the gas temperaturefrom the widths is difficult in this case because we do notobserve the whole line profile (our widths are measured atlower than half maximum) and we are not observing at diskcenter. In any case, assuming that width increases are due totemperature alone, using the relations of Leenaarts et al. (2012)we find that the typical width enhancements in spicules( mW W1.05 1.2( ) ) amount to temperature increases ofabout 500–1500 K compared to where Hα is formed alonglines of sight that do not intersect spicular material. A signatureof increased line widths in spicules has also been found in linesformed in higher temperatures, from the upper chromosphere tothe transition region (Pereira et al. 2014; Tian et al. 2014;Rouppe van der Voort et al. 2015).With width maps as a robust method for detecting spicules,

we find that virtually all cases of the “suddenly appearing”spicules of JCR12 seen in narrowband images can be explainedby Doppler motions. This view is also corroborated by Shetyeet al. (2016) using a different set of observations, and was alsoproposed by Kuridze et al. (2015). JCR12 claim that only anoptical superposition effect (i.e., spicules as sheets) can explainthese “suddenly appearing” spicules. We find that such eventsonly appear suddenly when oneʼs observations are limited tosingle wavelength narrowband images—using the completespectral information or width maps we find that the structurewas already there, before suddenly appearing at a particularwavelength. Limb spicules have transverse motions on the

Figure 5. Hα observations of spicules at the limb. From top to bottom, panels show the time evolution in the Hα blue wing intensities at −23 km s−1 (−0.05 nm)from line center, the combined Dopplergram (see text), the Hα red wing intensities at 23 km s−1 (0.05 nm) from line center, and a filtergram with a 0.1 nm GaussianFWHM. The spatial size of this region is 8 × 12 Mm2. The time increases from left to right, with the time difference from the first frame shown on top. The startingUT time of the sequence is 10:29:53.

8

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson

order of 5–30 km s−1 (e.g., Pereira et al. 2012). Such motionsalong the line of sight (because there is no reason to assume itis a preferred direction) are enough to cause the changingDoppler shifts that in some cases make spicules appear anddisappear suddenly from a narrowband image, requiring noadditional abstraction of spicules as fluted sheets.

6. CONCLUSIONS

We make use of a unique set of high spatial and temporalresolution observations in several wavelengths of Hα and Ca II

H filtergrams, obtained during a period of remarkably good,stable seeing. We observe quiet Sun spicules inside the diskand off the limb, and our main findings can be summarized asfollows:

1. Spicules have the same properties in Hα and Ca II Hfiltergrams.

2. Spicules become more abundant in wavelengths closer tothe Hα line core, reflecting a distribution of Dopplershifts and widths.

3. Measuring the properties of spicules using “singlewavelength” narrowband filtergrams can be misleadingbecause Doppler shifts can make the spicule appear anddisappear prematurely.

4. Hα width maps provide a robust way of following theevolution of spicules.

5. The sudden appearance of spicules in narrowband imagescan be explained by transverse motions along the line ofsight; one does not need to evoke the sheet model assuggested by JCR12.

This research was supported by the Research Council ofNorway through the grant “Solar Atmospheric Modelling” andby grants of computing time from the Programme forSupercomputing, and by the European Research Council underthe European Unionʼs Seventh Framework Programme (FP7/2007-2013) / ERC Grant agreement No. 291058. This workhas benefited from discussions at the International SpaceScience Institute (ISSI) meeting on “Heating of the magnetizedchromosphere” from 2015 January 5 to 8, where many aspectsof this paper were discussed with other colleagues. TheSwedish 1 m Solar Telescope is operated on the island of LaPalma by the Institute for Solar Physics of StockholmUniversity in the Spanish Observatorio del Roque de losMuchachos of the Instituto de Astrofísica de Canarias. Thisresearch has made use of SunPy, an open-source and freecommunity-developed solar data analysis package written inPython (SunPy Community et al. 2015).

REFERENCES

Beck, C., Rezaei, R., & Puschmann, K. G. 2013, A&A, 549, A24Beckers, J. M. 1968, SoPh, 3, 367

Beckers, J. M. 1972, ARA&A, 10, 73Carlsson, M., Hansteen, V. H., de Pontieu, B., et al. 2007, PASJ, 59, S663Cauzzi, G., Reardon, K., Rutten, R. J., Tritschler, A., & Uitenbroek, H. 2009,

A&A, 503, 577de la Cruz Rodríguez, J., Löfdahl, M. G., Sütterlin, P., Hillberg, T., &

Rouppe van der Voort, L. 2015, A&A, 573, A40De Pontieu, B., Erdélyi, R., & James, S. P. 2004, Natur, 430, 536De Pontieu, B., McIntosh, S., Hansteen, V. H., et al. 2007, PASJ, 59, 655De Pontieu, B., McIntosh, S. W., Carlsson, M., et al. 2011, Sci, 331, 55De Pontieu, B., McIntosh, S. W., Hansteen, V. H., & Schrijver, C. J. 2009,

ApJL, 701, L1De Pontieu, B., Title, A. M., Lemen, J. R., et al. 2014, SoPh, 289, 2733Grossmann-Doerth, U., & Schmidt, W. 1992, A&A, 264, 236Henriques, V. M. J. 2012, A&A, 548, A114Judge, P. G., & Carlsson, M. 2010, ApJ, 719, 469Judge, P. G., Reardon, K., & Cauzzi, G. 2012, ApJL, 755, L11Judge, P. G., Tritschler, A., & Chye Low, B. 2011, ApJL, 730, L4Kuridze, D., Henriques, V., Mathioudakis, M., et al. 2015, ApJ, 802, 26Langangen, Ø., De Pontieu, B., Carlsson, M., et al. 2008, ApJL, 679, L167Leenaarts, J., Carlsson, M., & Rouppe van der Voort, L. 2012, ApJ, 749, 136Lemen, J. R., Title, A. M., Akin, D. J., et al. 2012, SoPh, 275, 17Lipartito, I., Judge, P. G., Reardon, K., & Cauzzi, G. 2014, ApJ, 785, 109Lippincott, S. L. 1957, SCoA, 2, 15Nishikawa, T. 1988, PASJ, 40, 613Pasachoff, J. M., Jacobson, W. A., & Sterling, A. C. 2009, SoPh, 260, 59Pereira, T. M. D., De Pontieu, B., & Carlsson, M. 2012, ApJ, 759, 18Pereira, T. M. D., De Pontieu, B., & Carlsson, M. 2013, ApJ, 764, 69Pereira, T. M. D., De Pontieu, B., Carlsson, M., et al. 2014, ApJL, 792, L15Pneuman, G. W., & Kopp, R. A. 1977, A&A, 55, 305Pneuman, G. W., & Kopp, R. A. 1978, SoPh, 57, 49Roberts, W. O. 1945, ApJ, 101, 136Rouppe van der Voort, L., De Pontieu, B., Pereira, T. M. D., Carlsson, M., &

Hansteen, V. 2015, ApJL, 799, L3Rouppe van der Voort, L., Leenaarts, J., de Pontieu, B., Carlsson, M., &

Vissers, G. 2009, ApJ, 705, 272Rutten, R. J. 2006, in ASP Conf. Ser. 354, Solar MHD Theory and

Observations: A High Spatial Resolution Perspective, ed. J. Leibacher,R. F. Stein, & H. Uitenbroek (San Francisco, CA: ASP), 276

Rutten, R. J. 2012, RSPTA, 370, 3129Scharmer, G. B., Bjelksjo, K., Korhonen, T. K., Lindberg, B., & Petterson, B.

2003, Proc. SPIE, 4853, 341Scharmer, G. B., Narayan, G., Hillberg, T., et al. 2008, ApJL, 689, L69Sekse, D. H., Rouppe van der Voort, L., & De Pontieu, B. 2012, ApJ,

752, 108Sekse, D. H., Rouppe van der Voort, L., & De Pontieu, B. 2013a, ApJ,

764, 164Sekse, D. H., Rouppe van der Voort, L., De Pontieu, B., & Scullion, E. 2013b,

ApJ, 769, 44Shetye, J., Doyle, J. G., Scullion, E., et al. 2016, A&A, 589, A3Skogsrud, H., Rouppe van der Voort, L., De Pontieu, B., & Pereira, T. M. D.

2015, ApJ, 806, 170Sterling, A. C. 2000, SoPh, 196, 79Suematsu, Y., Wang, H., & Zirin, H. 1995, ApJ, 450, 411SunPy Community, T., Mumford, S. J., Christie, S., et al. 2015, CS&D, 8,

014009Tian, H., DeLuca, E. E., Cranmer, S. R., et al. 2014, Sci, 346, 1255711Tsiropoula, G., Alissandrakis, C. E., & Schmieder, B. 1994, A&A, 290, 285Tsiropoula, G., Tziotziou, K., Kontogiannis, I., et al. 2012, SSRv, 169, 181van Noort, M., Rouppe van der Voort, L., & Löfdahl, M. G. 2005, SoPh,

228, 191Vissers, G., & Rouppe van der Voort, L. 2012, ApJ, 750, 22Withbroe, G. L. 1983, ApJ, 267, 825Yurchyshyn, V., Abramenko, V., & Goode, P. 2013, ApJ, 767, 17Zhang, Y. Z., Shibata, K., Wang, J. X., et al. 2012, ApJ, 750, 16

9

The Astrophysical Journal, 824:65 (9pp), 2016 June 20 Pereira, van der Voort, & Carlsson